For so long I have not seen any useful reply to this post.!
The method which I am using is a script to edit ovf files. Here is how it goes.
Once the vm1.ova and vm2.ova files are present, convert them into .ovf format using ovftool. The script will start writing a new ovf(xml only) file. This file will have contents dumped from vm1.ovf and vm2.ovf xml files. Disk sections inside the xml file will point to same vmdk files. This new ovx xml file can again be converted into ova file using ovftool.
This method is tried manually once and it worked fine. The vApp was successfully deployed on the esx server.
The main advantage of using this method is that it doesnt require a running ESX host to generate the vApp.
Thoughts on this method anyone ?
Thanks,
Nikunj